A key step in the pathogenesis of M tuberculosis is its ability to cause caseating necrosis, parenchymal lung destruction, and cavity formation, which develop into the characteristic necrotizing bronchointerstitial pneumonia and bronchiolitis of tuberculosis. M tuberculosis is cytotoxic to epithelial cells in vitro, and we have shown that this cytotoxicity is associated with cell membrane permeation to lactose dehydrogenase and is mediated by necrosis of lung epithelial cells after infection with virulent mycobacteria (Dobos, K. M., Quinn, F. D. and King, C. H. 2000, Infect. Immun. 68:6300-6310). Interestingly, the attenuated M bovis BCG does not induce necrosis in this epithelial cell model suggesting that necrosis is related to the virulence of mycobacteria. Our working hypothesis is that M tuberculosis possesses factors that cause necrosis. We intend to identify the genes that encode or synthesize such factors and determine their functions. Towards this goal, we have been successful in isolating two such (necrosis-deficient) mutants with insertions into genes that have no known function by screening a transposon library of the Erdman strain of M tuberculosis (TN5370) for mutants that have lost their ability to cause cell membrane permeation and necrosis. Both nec mutants possess extremely interesting phenotypes when grown in mice. The first mutant (necA) appears to be highly attenuated for growth and virulence in SCID mice. This is an important result as it suggests that we have identified a gene whose product either directly causes necrosis or induces necrosis and thus should enhance our understanding of tuberculosis pathogenesis. Interestingly, the second mutant (necB) appears to kill SCID mice more rapidly than the parental strain. We intend to characterize these mutants, characterize the functions of the gene products, and extend this mutant isolation strategy to identify a large battery of mutants defective for necrosis of host cells.